Abstract

A layer technique is introduced in this paper to examine agent-based systems from three different levels of abstraction. The technique provides analysis capabilities of an agent system from the point of view of roles, agents, and corresponding implementation objects. The core of the technique involves the specification of meta-models for each level. Three different translators are implemented to interpret each model and to create a corresponding specification of the next subsequent level. The final translator generates code suited for a specific agent environment.

Keywords

Abstraction levels, meta-models, roles, agents, objects, model translations